Posted by Jeff C from dial128.w-link.net (208.151.227.15) on March 08, 1998 at 19:48:34:
In Reply to: Some comments posted by Chris Lundie on March 08, 1998 at 19:21:44:
>Environment Variables
>On a Voodoo there are lots of environment variables you can use to tweak the card. We could make use of some of them
>in Big RRed.
Definitely, should be easy to get the variables. What to do with them is different story.
>Antialiasing
>I don't think antialiasing is working right. In RRedline there has to be a very specific drawing state set before you can do
>antialiasing. I tried the Glide antialiased lines test and the lines got fatter, but I don't think they were smoothed out.
remember they are antialiasing in 3d, not 2d. Unfortunately, we are not sure what it supposed to look like. i know it is supposed
to have fog, too, but that doesn't work. Maybe we can find someone who has a 3dfx card to take screenshots for us. Then we'll know.
I have a stragne problem in that the lines are only up in the upper left quadrant. Is anyone else seeing this?
>Swap Interval
>If you set it to zero there might be problems. The Verite could swap buffers before the command buffer is done. If you set
>gl_swapinterval 0 in Quake 2, you will see really bad looking flashing artifacts, not just "tearing". I think that might happen
>here... Not sure. Maybe the user could have the choice to force the swap interval to 1.
The idea here is that if the application specifically asks for it not to wait, it won't. If glide asks it to wait, it will.
The non-interleaved buffers are fine by me. I'm not quite sure why you woul dwant that anyway.
I think the best optimizations we could do for planar polygons would be to figure out which plane they are on, so we didn't have
to do the float->fix each time for that plane, but that might be jsut as slow trying to figure out whcih plane we're on.
Actually, right now I am playing with the Zbuffer. I think I managed to get it working better, but I havent tweaked any test programs
to make them better yet. I think it actually has to do with grConstantColorValue, which is what I am playing with right now.
>Antialiased Points
>I think you are supposed to use VL_Particle or something like that. Although I'm not sure if it really does any antialiasing.
>It's pretty silly to antialias a point anyway. Basically it just makes a little blurry dot...
I thought I did those right. i do agree it is pretty silly though.
>Antialiased Polygons
>You have to draw the polygon, then set the drawing state for antialiasing, then draw the antialiased edges. Then set the
>drawing state back to what it was before... If there is a Z buffer you could ignore this because a real Voodoo can't do
>depth buffering and antialiasing at the same time. (Who knows why?)
I also thought I did this right, check the code for VL_AADrawTriangle. I think you have to AA each edge. strange.
>grGlideGetVersion
>you seem a bit confused. Here is the code:
>/* jeffc 030798 */
>/* hmmmm... */
I just thought that you could have done a strcpy instead, but didn't bother to change it. Does it have to be 80 characters, or
can it be less?
I am still plugging away with some things, especially grConstantColorVariable and a few others.
There might even be another release tonight.
Sorry Chris, I guess I should have mailed you a copy right away, but it was 3:00am, and I dind't think about it.